Re: svn commit: r1792169 - in /httpd/httpd/trunk: CHANGES include/ap_mmn.h include/httpd.h modules/generators/mod_status.c modules/proxy/mod_proxy.c server/config.c server/util.c

[Eric Covener]

The rewrite case was failing in the test suite. I removed both checks
in r1792169.

On Mon, May 8, 2017 at 8:04 PM, Eric Covener  wrote:
> On Thu, Apr 27, 2017 at 1:51 PM, Eric Covener  wrote:
>> On Fri, Apr 21, 2017 at 4:44 AM,   wrote:
>>> +/* A request that has passed through .htaccess has no business
>>> + * landing up here.
>>> + */
>>> +if (ap_request_tainted(r, AP_TAINT_HTACCESS)) {
>>> +return DECLINED;
>>> +}
>>> +
>>
>> If AllowOverride is enabled for the document root an d an htaccess is
>> present,  this renders /server-status unreachable, regardless of
>> what's in the htaccess. If we're going to block this by default, we
>> might as well just stop configuring it with SetHandler and then the
>> taint checking is not needed.
>>
>> We also have in another thread the issue with RewriteRule ... [P] in
>> htaccess being blocked.  We need some kind of way to express a policy
>> that will be unique to different handlers.
>
> bump? Right now the only two protected handlers are blocking pretty
> routine configurations.



-- 
Eric Covener
cove...@gmail.com

Re: New warning with gcc 7.1

[Ruediger Pluem]

On 05/26/2017 11:29 AM, Christophe JAILLET wrote:
> Hi,
> 
> when compiling with gcc 7.1, we have the following warning (in French):
> 
> util_expr_eval.c: Dans la fonction « ap_expr_eval_re_backref »:
> util_expr_eval.c:199:63: warning: comparaison entre un pointeur et la 
> constante caractère zéro [-Wpointer-compare]
>  if (!ctx->re_pmatch || !ctx->re_source || *ctx->re_source == '\0' ||
>^~
> util_expr_eval.c:199:47: note: vouliez-vous déréférencer le pointeur ?
>  if (!ctx->re_pmatch || !ctx->re_source || *ctx->re_source == '\0' ||
>^
> 
> (i.e. comparison between a pointer and 0)
> 
> 
> Knowing that we have:
> 
> typedef struct {
> ...
> /** the string corresponding to the re_pmatch */
> const char **re_source;
> ...
> } ap_expr_eval_ctx_t;
> 
> 
> Should the test in 'ap_expr_eval_re_backref' be something like:
> 
> if (!ctx->re_pmatch || !ctx->re_source || !*ctx->re_source ||
> **ctx->re_source == '\0' || ctx->re_nmatch < n + 1)

Looks sane to me as from a brief view it looks sane to me to assume that 
ctx->re_source is always != NULL at least if
ctx->re_pmatch != 0

Regards

Rüdiger

New warning with gcc 7.1

[Christophe JAILLET]

Hi,

when compiling with gcc 7.1, we have the following warning (in French):

util_expr_eval.c: Dans la fonction « ap_expr_eval_re_backref »:
util_expr_eval.c:199:63: warning: comparaison entre un pointeur et la 
constante caractère zéro [-Wpointer-compare]

 if (!ctx->re_pmatch || !ctx->re_source || *ctx->re_source == '\0' ||
   ^~
util_expr_eval.c:199:47: note: vouliez-vous déréférencer le pointeur ?
 if (!ctx->re_pmatch || !ctx->re_source || *ctx->re_source == '\0' ||
   ^

(i.e. comparison between a pointer and 0)


Knowing that we have:

typedef struct {
...
/** the string corresponding to the re_pmatch */
const char **re_source;
...
} ap_expr_eval_ctx_t;


Should the test in 'ap_expr_eval_re_backref' be something like:

if (!ctx->re_pmatch || !ctx->re_source || !*ctx->re_source ||
**ctx->re_source == '\0' || ctx->re_nmatch < n + 1)


Best regards,

CJ